Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Jon Hodgson]

Johnny B wrote on Tue, 04 October 2005 20:02

Emmm, the basic idea I'm trying to relate is that for every intermediate step you can eliminate, the less errors will be introduced and the faster the overall process should be. Just for the purposes of discussion, let's say you did everything in 64-bit words. Sure, there would be some bits wasted in some of the process, but doing away with some of the conversions in the chain may yield some unexpected benefits. It's not enough to simply attack new or novel approaches based solely on theoretical analysis, you actually have to try new things to see what will happen. Anyway, here's my major point...just look at the data flow and see how it might be improved...can we eliminate steps? What happens when we eliminate steps? I hope you are getting the idea I'm trying to express. Speed up the data flow and make it more efficient on an overall system basis.

Johnny,

Theoretical analysis?

Unlike you I am not guessing at these things, programming and optimizing DSP code on various processors is how I make my living. My experience is practical, not theoretical.

[Johnny B]

Jon,

Fine. So you see absolutely no areas that could be improved? You see absolutely no areas that should even be examined? Are you saying it's perfect? Maybe I'm misunderstanding you. That's easy for me to do, since I've been told by my wife that I have a small brain.

[Jon Hodgson]

Johnny B wrote on Wed, 05 October 2005 05:44

Jon, Fine. So you see absolutely no areas that could be improved? You see absolutely no areas that should even be examined? Are you saying it's perfect? Maybe I'm misunderstanding you. That's easy for me to do, since I've been told by my wife that I have a small brain.

Johnny,

Do you have selective memory, a reading impairment, or simply the attention span of a goldfish?

If none of those are the case, then how have you possibly missed the number of times I've stated on this forum that there IS room for improvement?

But some of us aren't just talking about these things, we're doing them every day, so we happen to know what we're talking about. This is not black magic, I know what the algorithms do, I know how the processors work, so when you drone on about analysing where performance improvements can be made by skipping steps due to processing in 64 bits.. I'VE ALREADY DONE THE ANALYSIS.


When Newton's equations of motion started being inadequate due to us doing things where relativity had an effect, can you imagine where we'd be if Einstein had listened to someone saying "Why don't you just step off the top of the building, you don't know, you might find that gravity doesn't work EVERY time"

[Johnny B]

Jon,

I have been told by some doctors that I suffer from ADHD, so I apologise for that defect.

Hmmm, IIRC, gravity laws work OK on Mother Earth, not so well in outer space. There are other areas that follow the same course.

I'm glad you feel there is room for improvement.

Given the trend towrd 64-bit, how about if you outline some of your improvement priorities for us?

[stuntbutt]

The pattern:

1) Johnny B starts thread with question

2) Someone with actual knowledge replies

3)  Johnny B implies person doesn't know what they are talking about


Why not try this:

1)  Start thread with question

2) Someone with actual knowledge replies

3) Thank poster for sharing

4) Incorporate new knowledge/insight into your own quest for understanding

-----------------------------------------------------------
John Katsafanas

[Johnny B]

I like his format better.

1. Start thread with question or link

2. Have debate or discussion

[Ronny]

Johnny B wrote on Wed, 05 October 2005 12:10

Jon, I have been told by some doctors that I suffer from ADHD, so I apologise for that defect. Hmmm, IIRC, gravity laws work OK on Mother Earth, not so well in outer space. There are other areas that follow the same course. I'm glad you feel there is room for improvement. Given the trend towrd 64-bit, how about if you outline some of your improvement priorities for us?

Gravity works the same in space as it does on earth, it's just that you don't notice it, in orbit you are still being pulled toward the earth by it's gravitational pull and the free fall gives the indication of no gravity. The larger an object is or actually the more mass it has, the more gravity it has. The farther away from the object, the less the gravity effects you. There is 240,000 miles of space between the earth and the moon and the earth is many times the mass of the moon, yet the moon still causes whole oceans to rise and fall several feet twice a day on the earth, a planet with much more mass. In reality, it is a physical law that works as far as we know, pretty much the same anywhere in the universe. There are variations, for example a black hole or star that has collapsed on itself that can bend and greatly effect lightwaves, but the physics of gravity are the same. It's one of the 4 major forces.  

[Johnny B]

Ronny, I do want to have this argument with about which kind of science works better in space and which works better on earth, however, I believe there is quite a bit of discussion going on about coming up with a "Unification Theory" and that "String Theory" is now in the spotlight. You may want to run some googles to learn more if you are that interested.


However, you may want to recall that the Bumble Bee was once thought to violate the laws of aerodynamics because its body weight was too heavy for its wing-span. And yet, this wonderful little creature still flies thru the air despite the old rules which said it was not possible.

The point is, and always has been, we need to keep pushing the research envelope if we ever want digital sound quality to advance.

Good scientists often admit their ignorance, bad scientists act like "know-it-all's."

[Jon Hodgson]

Johnny B wrote on Tue, 11 October 2005 17:05

Good scientists often admit their ignorance, bad scientists act like "know-it-all's."

And, pray tell, what would you call someone who acts like a "know-it-all", but doesn't even understand the science?

[Ronny]

Johnny B wrote on Tue, 11 October 2005 12:05

Ronny, I do want to have this argument with about which kind of science works better in space and which works better on earth, however, I believe there is quite a bit of discussion going on about coming up with a "Unification Theory" and that "String Theory" is now in the spotlight. You may want to run some googles to learn more if you are that interested. However, you may want to recall that the Bumble Bee was once thought to violate the laws of aerodynamics because its body weight was too heavy for its wing-span. And yet, this wonderful little creature still flies thru the air despite the old rules which said it was not possible. The point is, and always has been, we need to keep pushing the research envelope if we ever want digital sound quality to advance. Good scientists often admit their ignorance, bad scientists act like "know-it-all's."

I agree Johnny that we need to keep research going, but what you don't seem to understand is that it is going on as we speak. You act like all of the gear designers and digi R&D folks aren't trying to improve, when they "all" are. It's a continuous process and no one has given up on trying to improve. The problem lies with the area of improvement and the point of no returns. We've pretty much reached the limits of higher sample rates adding anything sonically and are close to the same with increased bit depths. What you are leaving out of the equation is the processing power and filtering that must be heightened when using higher sample rates. There is a point of diminishing returns and frankly most gear designers agree that 96k is sufficient. The only ones that don't are the ones that are selling 192k converters. Hat's off to Dan, he's a converter designer and he could play the same game and design 192k converters and market them, but fortunately for all of us, he's as honest as they come and rather than try to generate money with 192k ADC's, which he could no doubt design a top notch 192k converter, he's telling us that we have reached the point of diminishing returns at 96k. Actually IIRC, he said 60k was plenty, in the debate with Michal.

Johnny, I've run blind tests with digi consoles running at 44.1, 48, 88.2 and 96k, live mic and pre-recorded mic and instrument tracks. Particpants were accomplished musicians, a couple high profile producers and other engineer colleagues and friends of mine and not one person could consistently identify which sample rate was being used. That's between 44 and 96k and now you want to go to 384k. I'm quite confident that if you were to perform the same tests that I have, with experienced ears and long time industry participants, that your results would mirror mine and you would realize how futile it is to try to improve digital audio by raising the sample rates any higher than they are now. Any new improvements on converters, IMHO, will be on the analog sides of the converters, however how audible this will be, remains to be seen and heard. It's not that folks are giving up research, it's that they have already reached the point of diminshing returns with the higher sample rates and therefore research is going on more intensely in other areas.

[Johnny B]

Ronny I understand this position, truly I do.

Mind if I ask you a question? When you ran those tests, how many channels and instruments were you using?

I just don't think that digital sounds as good as it could. I think the CD and MP3 formats were poor design choices, and I can hear things in digital that I really don't like.

Right now, I still prefer the sound of analogue...so kill me.  

[Ronny]

Johnny B wrote on Wed, 12 October 2005 03:20

Ronny I understand this position, truly I do. Mind if I ask you a question? When you ran those tests, how many channels and instruments were you using? I just don't think that digital sounds as good as it could. I think the CD and MP3 formats were poor design choices, and I can hear things in digital that I really don't like. Right now, I still prefer the sound of analogue...so kill me.

Mics and instruments varied, however I used the same ADC's and DAC's and routed the ADC's to different channels in the digital domain, so there was no analog path or switches between the examples.  

[Johnny B]

Ronny,

Lemme clarify my question a bit, ok?

When you say you did some comparisons, did you just compare a single instrument recorded on a single channel, or did you compare 80 instruments recorded on 80 different channels. I know 80 is a big number, so maybe somewhere in the range of greater than one and less than 80...IOW, this is a single track vs a multi-track sort of question.

[Ronny]

Johnny B wrote on Wed, 12 October 2005 18:18

Ronny, Lemme clarify my question a bit, ok? When you say you did some comparisons, did you just compare a single instrument recorded on a single channel, or did you compare 80 instruments recorded on 80 different channels. I know 80 is a big number, so maybe somewhere in the range of greater than one and less than 80...IOW, this is a single rack vs a multi-track sort of question.

I've done several tests, but the one that I was referring to, just a few tracks, most of the live testing was 3 vocalists, but one vocalist at a time through a PA system. The recorded tracks were 3 mics on acoustic guitar and single vocalist track, same song recorded 4 times each at 44, 48, 88 and 96k and played back at each sample rate after all 4 examples were recorded. So no my tests didn't include what sampling rates do to 80 tracks, but over busy mixes aren't always condusive to these kind of tests where you are looking for subtle differences. My only regret is that I couldn't figure out a way to seemlessly A/B the singers on the live tests, because each wordclock change always took a few seconds.  

[Johnny B]

Well I'm not talking about overly busy mixes, merely ones with more than a few tracks.

Once the plug-ins are added to multi-track digital recordings is where I hear all sorts of problems.  You could blame it on bad code, I suppose, but I think there are some major underlying issues which make addressing these problems rather difficult with the current implementations of the currently popular technology.

In fact, if enough digital tracks are loaded up without any plugs whatsoever, I can still hear problems.  I think others hear these problems as well and have complained about it. Simply because they have not translated what they hear into precise scientific terms or math formulas, they have been subject to ridiclue and held up to scorn. I think this is grave mistake. Often the "ear people" and the musicians can hear things before the engineers get around to figuring out that these "ear people" are right about hearing the defects.

In my view, the "ear people" ought to be elevated in the process of moving digital technology forward. Although the descriptor "ear people" is somewhat of a misnomer because you are not limited to the people's ears, but rather, are looking at the extremely complex system that make up entire human beings, and that takes in the the body and the brain, among other things, the "ear people" moniker will have to do until a better descriptor comes along. In any event, this group should not be discounted and should, in fact, be elevated.

Perhaps another way to look at this would be to say, "When the 'ear people' are completely satisfied, then most human beings should be satisfied with the sound of digital. Satisfy the 'ear people,' and the analogue vs. digital debate should subside."